moderated Re: Modifying Pronunciations in Eloquence Software: Is it possible?


Annabelle Susan Morison
 

Hey Glenn!
What I want is for the words to be spoken as if they were one. "soninlaw", if that's the way you might be thinking of.

On 03/02/2022 8:23 AM Glenn / Lenny <glenn@...> wrote:


As pronunciation of punctuation goes, can't someone just increase or decrease the punctuation level in the settings?
I don't know if Annabelle wants the dashes spoken or not.
Like changing it from some to most or vice versa.
Glenn
----- Original Message -----
From: Leo
Sent: Wednesday, March 02, 2022 9:19 AM
Subject: Re: Modifying Pronunciations in Eloquence Software: Is it possible?

I don’t actually know if the Eloquence dictionary file does not take the hyphenated words  as lexical units because some kind of support must be added, or because Jaws is incapable of sending the hyphen to the synth, or even both.  As for other synthesizers such as vocalizer or Eloquence for Windows,  each one has their own native representation. The problem for the user is getting to know where to add those entries. This is why is really important that the companies behind the engines add this kind of support  for the custom user’s dictionaries.



Best,

Leo Bado


From: main@jfw.groups.io <main@jfw.groups.io> On Behalf Of Annabelle Susan Morison
Sent: Wednesday, March 2, 2022 8:59 AM
To: main@jfw.groups.io
Subject: Re: Modifying Pronunciations in Eloquence Software: Is it possible?


Yaaaaaaaaaaayyyyyyy! That time it actually worked! Now I'd like to find out how to do the ones for those "in-law" compounds. Would it be the same way as what you described in your previous email? Better yet, would it work with other synthesizers as well? Like Vocalizer, for example?

On 03/02/2022 6:38 AM Leo <literophilus@...> wrote:



I’m sorry, the format changed when I pasted the notations in the e-mail message. You can fix it easily though.

Open the dictionary file again,  and for each line delete the blank space between the lexical unit and the sign ` and insert a tabulation. Here is an example:

Incorrect:

Voila `[.0vwa.1la]

Correct:

Voila `[.0vwa.1la]


Do the same for every single entry, and don’t forget always opening the notepad as administrator and saving all changes.


Best,

Leo Bado


From: main@jfw.groups.io <main@jfw.groups.io> On Behalf Of Annabelle Susan Morison
Sent: Wednesday, March 2, 2022 4:57 AM
To: main@jfw.groups.io
Subject: Re: Modifying Pronunciations in Eloquence Software: Is it possible?


Hey Leo!

I tried your suggestion, note-for-note, and when I typed "voila", in a regular notepad file after saving the dictionary file as instructed, the pronunciation didn't change! What did I do wrong? Is there something I was forgetting?

On 03/01/2022 12:06 PM Leo <literophilus@...> wrote:



OK folks,

first and foremost, I want to respond to Annabelle since she brought up this matter.

I’m 99.9 percent sure that the Jaws dictionary manager does not support any SSML tag. My reasoning is quite simple even though I’m far, far away from being an expert or a developer. The documentation related to the synth states that, for the matter of defining the specific pronunciation of a word, either IPA or SPR translation  must be used. Now, here comes the key to understand the difference between orthographic and phonetic representation. Both IPA and SPR are systems that represent phonemes whereas orthographic representation is what the Jaws dictionary manager really supports, and this is the reality for most synthesizers nowadays because the companies behind these synthesizers either neglect users or consider users  incapable of dealing with this particular feature of the speech engines.

Now, let’s go back and have a real quick glimpse  into both IPA and SPR. IPA stands for international phonetic alphabet, and it is largely used by linguists, general dictionaries, and people learning languages  to check the standardized way of how a particular sound in any given language is represented.  On the flipside, SPR stands for symbolic phonetic representation, and it is specifically used by the IBM speech engine  as an built-in feature through which the synth interprets  how a particular word is pronounced.

Well, it happens to be that this same built-in feature, SPR, is present in all the engines derived from the IBM engine, such as via voice and of course, eloquence.

Now, before diving into how to add a new entry using SPR, I want to give you my 2 cent worth  on modifying  inflection, pitch, and even intonation.

As I told you, I’m not a developer, but I’ve read developer’s opinions in this regard indicating that this is only possible via using developer’s kits and tools, and knowing exactly what you’re doing since, particularly the aspects of inflection and intonation are not as simple as they look.  Of course, if you want the changes to be permanent. For any other purpose, you should ask the guys who use some tags to make Eloquence able to sing.

 

OK, let’s put the show truly on the road:

 

If you haven’t made any modification to the following words, you will notice how terribly such a words are pronounced by eloquence:

Voila

vernissage

Descartes

Sisyphean

 

Let’s create the necessary entries to teach this Eloquence guy how to pronounce standard American English.

 

For those who want to know how this works, you may want to follow these instructions:

•    In your computer, launch  the notepad app as administrator. (You can do it easily by pressing windows key plus r, then type “notepad” without the quotes, now push and hold control and shift at the same time, and push enter.)

 

• Press alt plus Y when the user account control dialogue pops up. Now, press control plus o to open a new file.

 

• Now press control plus l, which will take you to the address bar, and delete the current address. Paste the following address instead, and hit enter:

C:\Program Files (x86)\Freedom Scientific\Shared\Eloquence\6.1

 

• Presently, tab until you find the field “files of type,” and in this combo box choose “all files.”  Now in this same dialogue, get back to the explorer where the files are  shown, and look for the file called “ELOQMAIN,” and open it.

 

• Now go to the end of the document by pressing control plus end, and paste the following notations:

 

 

Voila `[.0vwa.1la]

VOILA `[.0vwa.1la]

voila `[.0vwa.1la]

Vernissage `[.2ver.0nX.1saZ]

VERNISSAGE `[.2ver.0nX.1saZ]

vernissage `[.2ver.0nX.1saZ]

Descartes `[.0de.1kart]

DESCARTES `[.0de.1kart]

descartes `[.0de.1kart]

Sisyphean `[.2sI.0sX.1fi.0Xn]

SISYPHEAN `[.2sI.0sX.1fi.0Xn]

sisyphean `[.2sI.0sX.1fi.0Xn]

 

End of notations.

 

Once you have pasted  the notations, save the document and exit. Come back to this document, and check again how the words in the beginning of the instructions are pronounced.

By this time you will have noticed that every word has three entries. This would not be the case if Jaws added  support for a root file and an abbreviation file.  The essential difference between a root dictionary file and a main dictionary file is that a root  dictionary file is non-case-sensitive.  Of course, Eloquence supports the three dictionary files, but it is just a real mystery why it haven’t been included.  Not only this, but according to the documentation available, the three dictionary files can be added for several languages supported, such as Spanish and German.

Back to the notations, those of you who are observant  will have noticed that the notations include three essential elements to define the pronunciation of a word, namely, syllable boundaries, primary and secondary stress, and the actual phoneme.

As for the example that Annabelle brought up,  namely, the pronunciation of son-in-law, there is a serious  difficulty, again, in my humble opinion, which would be the way how Jaws usually reads  compound words because Jaws doesn’t seem to interpret the dash symbol  in the middle of a text as a lexical unit; however, I know that I can be wrong.  It would be helpful, however, if someone knew how to make Jaws  send the compound words to the synth as a lexical unit.

 

If somebody is interested, There is someone who goes by the name Amir, and he has been building a dictionary file for long time, including hundreds of fixed words. So, let me know in case you want the file.

 

Also, the you can find the online documentation  in this link:

https://www.ibm.com/support/knowledgecenter/fr/SSMQSV_6.1.1/com.ibm.voicetools.ssml.doc/ssml_sprs_intro.html#ssml_sprs_intro

 

Last but not least, I wrote months ago to FS support requesting if they could make available more functionality  to improve the pronunciation of Eloquence, and I even explained everything that I have written here, but I never got a response.

I also wrote to CodeFactory, since I bought  years ago Eloquence for Windows SAPI 5,  asking for a way to add SAPI 5 phonemes to the user dictionary, but nothing has ever happened.

 


Best,

Leo Bado


From: main@jfw.groups.io <main@jfw.groups.io> On Behalf Of Brian Vogel
Sent: Monday, February 28, 2022 12:49 PM
To: main@jfw.groups.io
Subject: Re: Modifying Pronunciations in Eloquence Software: Is it possible?


On Mon, Feb 28, 2022 at 06:54 AM, Leo wrote:

But users don’t seem to care too much about it, partly because people don’t understand the difference between orthographic  and phonetic representation.

-
Which is understandable, after all.  But, that being said, those users can use the "Mute this topic" feature.

How to get maximum possible control over pronunciation in a given synth is something I'd have to believe a certain subset on any screen reader group would have keen interest in.  And those who don't have an easy way to silence the topic itself.

I'll post my tutorial again, as it never hurts for the newer members on high-traffic groups to be aware of their options with the Groups.io service:  

Controlling the Messages You Receive via E-Mail from Groups.io (docx)

Controlling the Messages You Receive via E-Mail from Groups.io  (PDF)


--

Brian - Windows 10, 64-Bit, Version 21H2, Build 19044  

Friendship. . . Intimacy, untroubled by eros.

         ~ Kim Chernin




Join main@jfw.groups.io to automatically receive all group messages.